Content uploaded by Yue-Hua Sun
Author content
All content in this area was uploaded by Yue-Hua Sun on Jun 11, 2016
Content may be subject to copyright.
BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors, nonprofit publishers, academic institutions,
research libraries, and research funders in the common goal of maximizing access to critical research.
Predicting threat of climate change to the Chinese grouse on the Qinghai—Tibet
plateau
Author(s): Nan Lyu and Yue-Hua Sun
Source: Wildlife Biology, 20(2):73-82.
Published By: Nordic Board for Wildlife Research
DOI: http://dx.doi.org/10.2981/wlb.13024
URL: http://www.bioone.org/doi/full/10.2981/wlb.13024
BioOne (www.bioone.org) is a nonprofit, online aggregation of core research in the biological, ecological,
and environmental sciences. BioOne provides a sustainable online platform for over 170 journals and books
published by nonprofit societies, associations, museums, institutions, and presses.
Your use of this PDF, the BioOne Web site, and all posted and associated content indicates your acceptance of
BioOne’s Terms of Use, available at www.bioone.org/page/terms_of_use.
Usage of BioOne content is strictly limited to personal, educational, and non-commercial use. Commercial
inquiries or rights and permissions requests should be directed to the individual publisher as copyright holder.
73
Predicting threat of climate change to the Chinese grouse on the
Qinghai–Tibet plateau
Nan Lyu and Yue-Hua Sun
N. Lyu and Y.-H. Sun (sunyh@ioz.ac.cn), Key Laboratory of Animal Ecology and Conservation Biology, Inst. of Zoology, Chinese Academy of
Sciences, CN-100101 Beijing, PR China
e Chinese grouse Tetrastes sewerzowi is restricted to small mountain areas on the southeastern edge of Qinghai–Tibet
plateau. Recent evidence indicates that the global climate has undergone rapid change. To assess the potential effects
of climate change on Chinese grouse, we applied a maximum-entropy modeling (MaxEnt) method to predict the
current and future distributions of this species for three time periods: 2020, 2050 and 2080 in two greenhouse-gas
emissions scenarios (A2a and B2a), which assume a medium and a lower increase in CO2 emissions, respectively. Our
modeling revealed that: 1) the size of suitable areas for grouse will decline over time, especially in emissions scenario
A2a; 2) range shifts will happen at both latitudinal (northward shift) and elevational direction (upward). In addition,
habitat expansion will be limited relative to loss, especially in the more distant future. Although the size of suitable
area will not change greatly in the near future (e.g. 2020 and 2050), as predicted in the emissions scenario A2a in
2020, habitat will become more fragmented. erefore, we suggest that the habitat fragmentation be considered with
range shifts calculation while assessing the climate change threats. To cope with the ongoing climate change, either
the protected area of the existing reserves should be expanded or new reserves should be established to accommo-
date range shifts. Reforestation and gouse population monitoring should also be conducted in the reserves to track
response of grouse to climate change.
Recent evidence indicates that the global climate is undergo-
ing rapid change, and is predicted to continue over the next
century (Easterling et al. 1997, Leech and Crick 2007, IPCC
2007, Solomon et al. 2007). It has long been recognized that
the contemporary climate is related to species distributions
(Hawkins et al. 2003). erefore, climate change may pose
a threat to species by affecting population dynamics (Waite
and Strickland 2006), distributions (Hughes 2000, Walther
et al. 2002, Zimbres et al. 2012) and the spatial struture
of the suitable habitats (i.e. fragmentation, Lu et al. 2012a,
b). ese threats could result in local population extinction
(omas et al. 2004), especially for species having narrow
thermal tolerance (Forero-Medina et al. 2011), such as the
montane species (Pounds et al. 1999) and species living in
northern latitudes, like polar bears Ursus maritimus (Hunter
et al. 2010).
All grouse species only occur within the temperate, boreal
and Arctic biogeographical zones of the northern hemisphere
(Johnsgard 1983). ey are adapted to cold climates, as indi-
cated by feathered feet and nostrils and long intestines with
well-developed caeca that enable them to digest coarse win-
ter foods (e.g. buds and conifer needles; Johnsgard 1983).
Because they are adapted to cold climates, it is hypothesized
that they evolved in northern latitudes (del Hoyo et al. 1994).
Among them, the Chinese grouse Tetrastes sewerzowi is a rare
species (Storch 2000, Sun et al. 2003) and has been listed as
Near reatened with decreasing trend by IUCN (IUCN
2012). It is endemic and restricted to conifer-dominated
forests with deciduous trees on the southeastern edge of
Qinghai–Tibet plateau (Sun et al. 2003). In many areas, only
the wetter northern slopes have forest vegetation that sup-
port grouse populations (Sun et al. 2003), which results in
fragmentation of its habitat. Furthermore, large-scale defor-
estation and intensive livestock grazing exacerbates habitat
loss and fragmentation (BirdLife International 2001, Klaus
et al. 2009). erefore, we hypotheize that climate change
threatens this species not only by influencing range shifts,
but also by increasing habitat loss and fragmentation.
ere are many species distribution modelling (SDM)
methods to predict the potential distributions and estimate
climate change effects through projection of future spe-
cies distributions (Pearson and Dawson 2003, Guisan and
uiller 2005, Elith et al. 2006, Elith and Leathwick 2009).
By comparing predicted distributions under current con-
ditions to future climate change scenarios, managers can
develop more effective and suitable conservation plans
(Araújo and Rahbek 2006). erefore, we use the maximum-
entropy SDM method (Phillips et al. 2006) for scenarios
relavent to 2020, 2050 and 2080 to predict the potential cli-
mate change threats to Chinese grouse as a way to enchance
Wildlife Biology 20: 73–82, 2014
doi: 10.2981/wlb.13024
© 2014 e Authors. is is an Open Access article
Subject Editor: Ralph J. Gutiérrez. Accepted 14 October 2013
74
conservation planning. en we assess these threats from the
perspectives of range shifts and changing spatial structure
of suitable habitat. Climate change may entail habitat loss
and upward shifts of habitat, which could exacerbate habitat
fragmentation, and thus change the spatial structure (Lu et al.
2012a, b). Here we try to assess the habitat fragmentation
aspect by comparing some specific landscape metrics.
Methods
We constructed SDMs for Chinese grouse Tetrastes
sewerzowi within China. We used the same set of grouse
location (observation) records as those used by Lu et al.
(2012a). We reduced spatial autocorrelation (Dormann
et al. 2007) by setting a spatial distance threshold of 0.083
decimal degrees (i.e. 5a, about 8 km) between location
records. e location record was included only if the spatial
distances with all other location records was larger than the
threshold. Using these filtering criteria, we had 41 location
records, which we used to construct the SDMs.
Environmental predictors
To construct SDMs for the Chinese grouse, we collected 22
environmental predictors belonging to two different catego-
ries (bio-climatic and topographic, Supplementary material
Appendix l Table A1). We obtained 19 bio-climatic predic-
tors from the WorldClim database (Hijmans et al. 2005),
which have been used to predict distributions of other spe-
cies (Peterson et al. 2006, Cordellier and Pfenninger 2009,
Lu et al. 2012a). We also used slope, aspect and compound
topographic index, downloaded from the USGS database
(http://eros.usgs.gov), as topographic predictors. For
projecting future distributions, we extracted the same bio-
climatic predictors for 2020, 2050 and 2080 from three
internationally recognized general circulation models
(GCMs) (CCCMA, CSIRO and HadCM3) with two IPCC
greenhouse gas emissions scenarios (A2a and B2a) (IPCC
2001). According to IPCC Special Report on emissions
scenarios, the emissions scenarios cover a wide range of
assumptions about main driving forces of future emissions,
from demographic to technological and economic develo-
pments (Nakicenovic and Swart 2000). As suggested by
McKenney et al. (2007), we selected the A2a and B2a, which
reflected two climate scenarios based on a conservative and
an extreme emissions scenario in the future. e A2a sce-
nario was projected for less environmentally conscious, and
less regionalized solutions to economic, social, and environ-
mental sustainability than the B2a scenario (Zhang and Liu
2005, Hu and Jiang 2011, Araújo et al. 2011). erefore,
the A2a scenario described a world with higher population
growth rate, more rapid economic growth, faster land-use
changes and unreduced emissions, while the B2a described
a world with a reduced emission scenario where resource
conservation would be promoted in the early decades of this
century and the CO2 emissions would decline by midcen-
tury (Nakicenovic and Swart 2000, Solomon et al. 2007).
ese data were available from the WorldClim database
and the International Centre for Tropical Agriculture data-
base (Hijmans et al. 2005, Ramirez and Jarvis 2008). We
transformed all environmental predictors into a moderated
resolution of 5a (^8 8 km2) to construct the SDMs for the
Chinese grouse.
SDM construction and evaluation
We modeled the distribution of Chinese grouse using
MaxEnt (ver. 3.3.3k, Phillips et al. 2006). is model was
a machine-learning algorithm for predicting the species
potential distributions by combining the environmen-
tal predictors and the location data as input (Elith et al.
2011). MaxEnt used the location-only data and was appro-
priate for species with small samples of species locations
(Elith et al. 2006, Pearson et al. 2007, Wisz et al. 2008).
We selected the logistic output with suitability values rang-
ing from 0 to 1, which represented the occurrence prob-
ability of target species (Phillips and Dudík 2008). Further
parameter settings include: convergence threshold (10–5),
regularization multiplier (1) and the maximum number of
iterations (500). We used cross-validation with five rep-
licates to assess the robustness of SDM (Fouquet et al.
2010, Lu et al. 2012b).e whole set of location records
were partitioned as 80% training data and 20% testing
data during each replicate. In order to decrease predictive
uncertainty, we used the ensemble forecasting approach as
suggested by Araújo and New (2007). We applied the basic
mathematical function of mean ensembles to calculate the
final logistic outputs (Marmion et al. 2009). Due to the
uncertainties of different GCMs, we calculated the mean
values of three projected suitability outputs (CCCMA,
CSIRO and HadCM3) for the following analysis, which
has become the most commonly used techniques and
been shown to yield robust predictions (Hole et al. 2009,
Marmion et al. 2009, Hu and Jiang 2011). During the
modeling process, we selected the jackknife analyses of the
regularized gain with training data to examine the impor-
tance of different predictors.
In order to evaluate the predictive performances of the
SDMs, we selected the area under the receiver operating
characteristic curve (AUC; Fielding and Bell 1997) to evalu-
ate the predictions using the training and test data firstly.
is measurement has been used extensively in the species
distribution modelling literature (Loiselle et al. 2010), and
has been considered the best practice for assessing SDM
accuracy (Pearce and Ferrier 2000, uiller 2003, Rushton
et al. 2004, Austin 2007). AUC values ranged from 0 to 1,
where a value of 1 indicated perfect (100%) discrimination
and a score of 0.5 indicated a model with discrimination
that is no better than random (Pearce and Ferrier 2000). To
test the reliability of the accuracy assessment, we also calcu-
lated Cohen’s kappa statistic (kappa; Cohen 1960) and true
skills statistics (TSS; Allouche et al. 2006) as suggested by
Mouton et al. (2010). Both kappa and TSS values ranged
from –1 to 1 where 1 indicated a perfect performance,
while values 0 indicated a performance no different to
random (Cohen 1960, Allouche et al. 2006). We generated
random pseudo-absences using the Random Points function
of the R package dismo (Elith et al. 2006, Hijmans et al.
2010) and then using the presence.absence.accuracy function
of the R package PresenceAbsence to calculate the kappa and
TSS values (Freeman and Moisen 2008a).
75
Climate change threats assessments
For comparing distributions at future times, we projected
species distribution for 2020, 2050 and 2080, respectively.
To explore the climate change threats to the Chinese grouse,
we calculated the range, position (i.e. range centroid) and
elevation differences. Generally, range shifts predicted to
occur from climate change may have three forms – shifting
by latitude, by longitude and by elevation (Forero-Medina
et al. 2011). e suitable area (i.e. the entire spatial area over
which a species might be found) should be identified with a
reasonable suitability threshold. We used the optimal.thresh-
olds function in the PresenceAbsence R package (Freeman
and Moisen 2008a) to calculate the suitable threshold and
selected Pred Prev Obs as the threshold following Freeman
and Moisen (2008b) and Lu et al. (2012a). Subsequently,
we extracted the altitude, longitude and latitude of each
grid throughout the suitable area and compared these val-
ues between current and future distributions. Furthermore,
we compared the habitat fragmentation between current
and predicted future climate change scenarios. Specifically,
we applied a square lattice to identify the continuous suit-
able patches for which we considered the target grid to be
connected with eight neighboring grids (Lu et al. 2012a).
After identifying the continuous patches predicted as suit-
able for Chinese grouse, we applied several landscape metrics
to assess habitat fragmentation quantitatively (McGarigal
and Marks 1995): suitable area size (SAS) by calculating the
predicted suitable grid number; area size of concave poly-
gon (ASP) which contains all suitable grids; proportion of
suitable areas (PSA); number of patches (NP), patch density
(PD, number of patches per 1000 km2), mean patch size
(MPS), patch size standard deviation (PSSD) and patch size
coefficient of variation (PSCV). All these calculations and
classifications for the present and future distributions were
performed using the program ArcGIS ver. 9.3.1 (Environ-
mental Systems Research Institute 2009).
Results
Modeling evaluation
e kappa and TSS values of all SDMs were larger than
0.6, while the AUC values were larger than 0.9 (Table 1),
which revealed that the SDMs performed well for the Chi-
nese grouse. Cross-validation also suggested that the models
were quite robust, because of the relatively high values of
accuracy measurements for both training data and test data
(Table 1). According to the jackknife test of variable impor-
tance, the mean temperature of driest quarter (BIO10), max
temperature of warmest month (BIO5), mean temperature
of wettest quarter (BIO8), slope, temperature seasonal-
ity (BIO4) and isothermality (BIO3) achieved the highest
gains when used in isolation, and thus appeared to have the
most useful niche requirement information for the Chinese
grouse.
Area size and range shifts
Based on the current suitability map of Chinese grouse
(Fig. 1), the distribution range was largely concentrated
at the southeastern edge of Qinghai–Tibet plateau. e
moutainous regions in southern Gansu, middle Sichuan and
eastern Tibet were predicted to be relatively more suitable for
this species (i.e. with redder color; Fig. 1). SDM predicted
that the suitable area size will decline over time, especially in
the A2a missions scenario (Fig. 2a). Although the suitable
area size changed moderately for 2020 ( 12.7% in A2a and
–9.5% in B2a) and 2050 (–13.9% in A2a and –1.3% in
B2a), there was a larger change of –55.9% and –23.4% for
2080 in A2a and B2a, respectively. Furthermore, the suit-
ability was predicted to decline for all climate change sce-
narios (i.e. the mean suitability will decline from 0.643 to
0.632, 0.620 and 0.586 for 2020, 2050 and 2080 in emis-
sions scenario A2a, and to 0.628, 0.624 and 0.608 in B2a,
respectively; see also Fig. 2b).
Habitat loss and gain in different climate change scenar-
ios during our timeframes indicated that horizontal range
shifts should occur. Suitable areas on the southern, western
and eastern margins of its current distribution would be lost,
while newly gained suitable areas would be located mainly
on the northern margin, especially in the more distant
future, (e.g. 2080, Fig. 3). In addtion, some newly gained
suitable areas in Tibet and western Sichuan in the near future
(e.g. 2020 or 2050) would be lost in the more distant future
(e.g. 2080; Fig. 3). According to the mean longitudes and
latitudes of suitable areas, we found that climate change
would require this species to shift northward generally
(Fig. 4a). Furthermore, although patterns of habitat loss and
gain were similar in different emissions scenarios, the magni-
tude of range change varied. In emissions scenario A2a, both
the stable and newly gained areas showed declining trends,
while areas lost areas should increase in the future (Fig. 4b).
We found that 82.2% of current suitable areas should be
stable in 2020, while this proportion should drop to 63.1%
and 30.6% in 2050 and 2080, respectively. Newly gained
areas were also predicted to decline from 30.5% in 2020 to
23.0% and 13.5% in 2050 and 2080, respectively. Range
change will be less dramatic in emmisssion scenario of B2a
(Fig. 4b). e stable areas will decrease from 73.2% in 2020
to 69.1% and 53.2% in 2050 and 2080, respectively. e
Table 1. Accuracy measurements of predictive SDMs for Chinese grouse. AUC area under relative operating characteristic curves; TSS true
skill statistic; Kappa Cohen’s kappa statistic; Rep.1–5 represent the five replicates for cross-validation.
Accuracy
measurements Ensemble
Rep.1 Rep. 2 Rep. 3 Rep. 4 Rep. 5
Training Test Training Test Training Test Training Test Training Test
TSS 0.893 0.941 0.933 0.894 0.800 0.921 0.762 0.914 0.838 0.906 0.864
Kappa 0.717 0.833 0.718 0.730 0.705 0.773 0.629 0.790 0.805 0.824 0.823
AUC 0.976 0.989 0.972 0.972 0.930 0.984 0.920 0.980 0.972 0.983 0.913
Ensemble represents the result of averaged five replicates.
76
According to the patch isolation maps (Fig. 5), both habitat
loss and upward shift exacerbated the fragmentation, espe-
cially in the A2a scenario. In the middle of Sichuan Province
(Fig. 5d–f), climate change would mainly compel Chinese
grouse to shift upward. A large number of suitable grids at
relatively lower altitudes would be lost, while newly gained
suitable grids would be at even higher altitudes. Suitable
habitat would become more isolated and perforated in the
future (Fig. 5d–f). e patch isolation patterns of the other-
timeframes can be found in the Supplementary material
Appendix 1 Fig. A1.
Discussion
Climate change threats
As the world’s largest geomorphology unit, the Qinghai–
Tibet plateau influences atmospheric circulation and affects
the climate of eastern Asia (Sun 1996). Coniferous forest
only occur in the mountainous areas at relatively higher alti-
tude on the southeastern edge of the Qinghai–Tibet plateau.
newly gained areas even showed an increasing trend from
17.3% in 2020 to 29.5% and 23.3% in 2050 and 2080,
respectively. Finally, from the perspective of vertical (i.e. ele-
vational) range shifts, the modelling results showed upward
shifting under climate change (Fig. 4c). e mean altitude
was predicted to increase from 3476 m to 3683 m, 3710 m
and 3848 m in 2020, 2050 and 2080, respectively, in emis-
sions scenario A2a, and from 3476 m to 3577 m, 3725 m
and 3741 m in 2020, 2050 and 2080, respectively, in emis-
sions scenario of B2a.
Habitat fragmentation
In general, climate change was predicted to affect grouse
through change in spatial structure of suitable areas. e
mean patch size will be reduced, while the patch density
will increase in both emissions scenarios and different time-
frames (Table 2). Moreover, we found that although the
value of patch size coefficient of variation would not change
greatly in the near future (e.g. 2020 and 2050) and would
even increase from 7.61 to 8.10 in 2020 in A2a, the mod-
elling showed declining trends in both emissions scenarios.
Figure 1. Mean predicted probability of occurrence (suitability) of the Sichuan jay under current situation, showing provinces in China.
e color represents the suitability, from low (green) to high (red).e right figure shows the area marked by red square in the left one,
and overlays with altitude (the whiter color indicates higher altitude). e dark blue dots are the occurrence records used for model
calibration.
Figure 2. Mean suitable area sizes (a) and suitabilities (b) of Chinese grouse under current and future climate change scenarios at three time
slices of 2020, 2050 and 2080 projected for the Qinghai–Tibet plateau. We use the number of suitable grids (8 8 km2 resolution) to
represent the size of suitable area. Current suitable grid number is illustrated using the dashed line in (a). e solid horizontal line represents
the median, short horizontal lines represent the maximum and minimum, the square symbol represents the mean, edges of box are
quartiles, whiskers are 5th and 95th percentiles and crosses are 1st and 99th percentiles.
77
Figure 3. Predicted potential Chinese grouse habitat that is stable, lost or gained under climate change, based on two IPCC greenhouse gas emission scenarios of A2a (a, b, c) and B2a (d, e, f) and
three time timeframes of 2020 (a, d), 2050 (b, e) and 2080 (c, f).
78
climate change may pose threats to Chinese grouse, and
these threats would likely intensify over time. omas and
Lennon (1999) proposed that the northern margins of
British birds has already extended by an average of 18.9 km
over a period of 20 years. Other studies have proposed that
the species on montane zones could shift to the higher ele-
vations (Pounds et al. 1999, Crick 2004). We found that
the Chinese grouse distributions would shift northward
and upward obviously under climate change scenarios (Fig. 3,
4). However, unlike species that have wide distributions or
greater dispersal capabilities, the availabilty of suitablilty pre-
dicted to expand under climate change would be limited for
Chinese grouse because of their lesser dispersal capability.
Moverover, it appeared that expansion of suitable habitat
ese forests formerly belong to the more widely distributed
Taiga forest, which has retracted to its present distributions
during the uplifting of the plateau (Cheng 1981). Due to
the unique natural ecological environment and geologi-
cal history, many endemic species have evolved and live in
these mountainous areas, such as the Sichuan jay Perisoreus
internigrans (Jing et al. 2009, Lu et al. 2012b), blood pheas-
ant Ithaginis cruentus (Jia et al. 2010), and snowy-cheeked
laughingthrush Garrulax sukatschewi (Wang et al. 2011).
Many of them have evolved specific survival strategies for
local climates, which could hinder adaptations necessary
during a regime of rapid climate change (Lu et al. 2012b).
Similar to other montane species (Sekercioglu et al.
2007, Wilson et al. 2007), our modelling revealed that
Figure 4. Climate change threats to Chinese grouse through range shifts. (a) the centroid of current and future suitable girds (through
calculating the mean longitude and latitude values); (b) the proportion of habitat that is stable, lost or gained when climate change; (c) the
mean altitude in current and climate change scenarios. e solid horizontal line represents the median, short horizontal lines represent the
maximum and minimum, the square symbol represents the mean, edges of box are quartiles, whiskers are 5th and 95th percentiles and
crosses are 1st and 99th percentiles.
Table 2. Quantitative habitat fragmentation assessment of the suitable areas for Chinese grouse. SAS suitable area size (km2); ASP area
size of concave polygon (km2); PSA proportion of suitable areas (%); NP number of patches; PD patch density (patches per 1000 km2);
MPS mean patch size (the number of grids); PSSD patch size standard deviation; PSCV patch size coefficient of variation (%).
Scenarios SASASP PSA NP PD MPS PSSD PSCV
Current 235060 821589 0.29 214 0.260 13.17 100.31 7.61
A2a 2020 264995 851896 0.31 274 0.322 11.60 87.14 7.51
2050 202374 765673 0.26 240 0.313 10.11 74.36 7.35
2080 103730 553421 0.19 197 0.356 6.31 25.91 4.10
B2a 2020 212713 827297 0.26 300 0.363 8.50 68.86 8.10
2050 231975 792484 0.29 248 0.313 11.22 81.87 7.30
2080 179943 714559 0.25 227 0.318 9.51 63.85 6.72
SAS is derived from the original calculated suitable grid number (multiplying 64 (the area of a grid) by grid number).
79
Figure 5. e isolation patterns of predicted suitable habitat for Chinese grouse. e isolated patches are indicated with different colors: (d), (e) and (f) shows the area marked by red square in (a), (b)
and (c), respectively. e figures of (d), (e) and (f) overlays with altitude (the wighter color indicates higher altitude).
80
plans. First, the protected area of the existing reserve for
Chinese grouse should be expanded or new reserves estab-
lished according to the current distributions and potential
for future range shifts. For example, only 23% mature forest
in the Lianhuashan Mountains was within the Lianhuashan
Nature Reserve (Sun et al. 2003). erefore, more surround-
ing forest area should be included within the reserve. Chi-
nese grouse have poor long-distance flying ability, thus, the
major portion of travel time during dispersal will be spent
walking, which is similar as ruffed grouse Bonasa umbellus
(Godfrey and Marshall 1969). More and larger reserves could
also improve the conservation efficiency through unified
planning, such as establishing movement corridors among
populations to facilitate dispersal maong habitat fragments.
Furthermore, human activities such as illegal hunting and
egg collecting still happens in local areas (Storch 2000). Egg
collecting is apparently substantial (Sun et al. 2003). Lu and
Sun (2011) even suggested that these activities may be the
most important factors that influence the local population
viability in Lianhuashan. Establishing or expanding reserves
could also improve the individual survival rate through ani-
mal protection (e.g. protecting individuals from illegal hunt-
ing and restrict human activity during the breeding season)
and community education about biodiversity conservation.
Second, reforestation should be conducted throughout the
mountain habitats of Chinese grouse, since rapid deforesta-
tion over the past century was largely due to humans rather
than the natural factors. Finally, grouse population monitor-
ing should be conducted, which would necessary to observe
population trends and assess the success of the conserva-
tion efforts (Storch 2000). Furthermore, attention should
be given not only to populations within the most suitable
and continuous area, but also to populations at the northern
and southern margins of the species’ distribution (Lu et al.
2012b) because populations at the margins of distributions
should be the most sensitive to the climate change.
Acknowledgements – is study was supported by grants of the
National Natural Science Foundation of China (31172099,
31071931). We sincerely thank the editor for their valuable com-
ments and corrections on the original version of the manuscript.
References
Allouche, O. et al. 2006. Assessing the accuracy of species distribu-
tion models: prevalence, kappa and the true skill statistic
(TSS). – J. Appl. Ecol. 43: 1223–1232.
Araújo, M. B. and Rahbek, C. 2006. – How does climate change
affect biodiversity? – Science 313: 1396–1397.
Araújo, M. B. and New, M. 2007. Ensemble forecasting of species
distributions. – Trends Ecol. Evol. 22: 42–47.
Araújo, M. B. et al. 2011. Climate change threatens European
conservation areas. – Ecol. Lett. 14: 484–492.
Austin, M. 2007. Species distribution models and ecological the-
ory: a critical assessment and some possible new approaches.
– Ecol. Modell. 200: 1–19.
BirdLife International 2001. reatened birds of Asia: the BirdLife
International Red Data Book. – BirdLife International, Cam-
bridge.
Cheng, T. 1981. e fauna and its evolution of territorial verte-
brates at Qinghai–Tibet plateau. – Rep. Beijing Nat. Mus. 9:
1–17.
areas would not compensate for the loss of current suitable
areas, especially under our distant future scenario (Fig. 4b).
Habitat fragmentation has long been recognized as one of
the major causes of global biodiversity loss and local extinc-
tion (Kruess and Tscharntke 1994). Because of the high
demands for farmland, timber and firewood by local people,
deforestation of Chinese grouse’s mountain habitats contin-
ued until 1998 when disastrous flooding happened there,
which motivated greater conservation of forests (Sun et al.
2003). According to our modelling, climate change may
exacerbate existing habitat fragmentation patterns (Table 2).
is fragmentation may reduce or halt gene flow or perhaps
increase the extinction risk of isolated populations (Lu et al.
2012b). Specifically, we found that fragmentation would
happen in both patch isolation and within patch fragmen-
tation (Lu et al. 2012a). e suitable area size would not
change much in the near future (e.g. 2020 and 2050) and was
even predicted to increase in emissions scenario A2a in 2020
(Fig. 2a). However, modelling predicted that habitat would
become more fragmented rapidly such that the patch density
would be increased in all climate change scenarios and time-
frames (Table 2). Furthremore, although the values of patch
size coefficient of variation were relatively stable in the near
future, they showed accelerated declining trends (Table 2).
Patch size coefficient of variation provided an index of patch
to patch size variation relative to the mean value (McGari-
gal and Marks 1995, Jacquemyn et al. 2002). erefore, the
declining values of patch size coefficient of variation partially
revealed that some currently connected large patches would
become isolated and then reduce the size variability among
patches. is conclusion was also observed in patch isolation
(Fig. 5) where the orginal big suitable patch shown in green
color (Fig. 5d) would be isolated into more small patches
(Fig. 5e–f). We also found that the patches would show
more internal fragmention. For example, the contour of big
patch with green color in Fig. 5d did not change much in the
B2a scenario, but the suitable habitat would become more
perforated (Fig. 5f), and thus could decrease both the pro-
portion of suitable areas and mean patch size. According to
the quantitative habitat fragmentation assessment, the mean
patch size would be reduced under all climate change sce-
narios and timeframes (Table 2). e proportion of suitable
areas would change moderately in 2020 and 2050 but would
decrease in 2080 in both scenarios (Table 2). It also showed
a declining trend in the A2a scenario (Table 2). Finally, our
modelling results validated the importance of a combinating
habitat fragmentation analysis with range shift calculations
when assessing the climate change threats using the SDM
method.
Conservation implications
Most grouse species are habitat specialists with fairly nar-
row habitat preferences, and thus are susceptible to habitat
changes (Storch 2000). e Chinese grouse is one of the
three grouse species listed as Near reatened by IUCN.
Our modelling suggests that although the climate change
would not have immediate threats in the near future, the
potential for negative effects would increase over time.
erefore, conservation managers should consider the
effects of climate change when developing conservation
81
IPCC 2001. Intergovernmental panel on climate change third
assessment report – climate change 2001. – IPCC, Geneva,
Switzerland.
IPCC 2007. Climate change 2007: synthesis report. Summary for
policy makers. – www.ipcc.ch/pdf/assessment-report/ar4/
syr/ar4_syr_spm.pdf
IUCN 2012. IUCN red list of threatened species. Ver. 2012.2. –
www.iucnredlist.org.
Jacquemyn, H. et al. 2002. Patch occupancy, population size and
reproductive success of a forest herb (Primula elatior) in a frag-
mented landscape. – Oecologia 130: 617–625.
Jia, C. X. et al. 2010. Unusual incubation behavior and embryonic
tolerance of hypothermia by the blood pheasant (Ithaginis
cruentus). – Auk 127: 926–931.
Jing, Y. et al. 2009. Alloparenting in the rare Sichuan jay (Perisoreus
internigrans). – Condor 111: 662–667.
Johnsgard, P. A. 1983. e grouse of the world. – Univ. of Nebraska
Press.
Klaus, S. et al. 2009. Autumn territoriality of Chinese grouse
Bonasa sewerzowi at Lianhuashan Natural Reserve, Gansu,
China. – Int. J. Galliformes Conserv. 1: 44–48.
Kruess, A. and Tscharntke, T. 1994. Habitat fragmentation,
species loss, and biological control. – Science 264:
1581–1584.
Leech, D. and Crick, H. 2007. Influence of climate change on the
abundance, distribution and phenology of woodland bird spe-
cies in temperate regions. – Ibis 149: 128–145.
Loiselle, B. A. et al. 2010. Assessing the impact of deforestation
and climate change on the range size and environmental niche
of bird species in the Atlantic forests, Brazil. – J. Biogeogr. 37:
1288–1301.
Lu, N. and Sun, Y.-H. 2011. Population viability analysis and con-
servation of Chinese grouse Bonasa sewerzowi in Lianhuashan
Nature Reserve, northwest China. – Bird Conserv. Int. 21:
49–58.
Lu, N. et al. 2012a. Species-specific habitat fragmentation assess-
ment, considering the ecological niche requirements and dis-
persal capability. – Biol. Conserv. 152: 102–109.
Lu, N. et al. 2012b. Assessing the distributions and potential risks
from climate change for the Sichuan jay (Perisoreus interni-
grans). – Condor 114: 365–376.
Marmion, M. et al. 2009. Evaluation of consensus methods in
predictive species distribution modelling. – Divers. Distrib.
15: 59–69.
McGarigal, K. and Marks, B. J. 1995. Spatial pattern analysis pro-
gram for quantifying landscape structure. – Gen. Tech. Rep.
PNW-GTR-351. US Dept Agric. For. Serv., Pacific Northwest
Res. Stn.
McKenney, D. W. et al. 2007. Potential impacts of climate change
on the distribution of North American trees. – Bioscience 57:
939–948.
Mouton, A. M. et al. 2010. Ecological relevance of performance
criteria for species distribution models. – Ecol. Modell. 221:
1995–2002.
Nakicenovic, N. and Swart, R. 2000. IPCC special report on emis-
sions scenarios. – Cambridge Univ. Press.
Pearce, J. and Ferrier, S. 2000. Evaluating the predictive perform-
ance of habitat models developed using logistic regression. –
Ecol. Modell. 133: 225–245.
Pearson, R. G. and Dawson, T. P. 2003. Predicting the impacts
of climate change on the distribution of species: are
bioclimate envelope models useful? – Global Ecol. Biogeogr.
12: 361–371.
Pearson, R. G. et al. 2007. Predicting species distributions from
small numbers of occurrence records: a test case using cryptic
geckos in Madagascar. – J. Biogeogr. 34: 102–117.
Phillips, S. J. and Dudík, M. 2008. Modeling of species distribu-
tions with Maxent: new extensions and a comprehensive
evaluation. – Ecography 31: 161–175.
Cohen, J. 1960. A coefficient of agreement for nominal scales.
– Educ. Psychol. Meas. 20: 37–46.
Cordellier, M. and Pfenninger, M. 2009. Inferring the past to pre-
dict the future: climate modelling predictions and phylogeog-
raphy for the freshwater gastropod Radix balthica (Pulmonata,
Basommatophora). – Mol. Ecol. 18: 534–544.
Crick, H. Q. P. 2004. e impact of climate change on birds. – Ibis
146: 48–56.
del Hoyo, J. et al. 1994. Handbook of the birds of the World. New
World vultures to guineafowl (Vol. 2). – Lynx Editions.
Dormann, C. F. et al. 2007. Methods to account for spatial auto-
correlation in the analysis of species distributional data: a
review. – Ecography 30: 609–628.
Easterling, D. R. et al. 1997. Maximum and minimum tempera-
ture trends for the globe. – Science 277: 364–367.
Elith, J. and Leathwick, J. R. 2009. Species distribution models:
ecological explanation and prediction across space and time.
– Annu. Rev. Ecol. Evol. Syst. 40: 677–697.
Elith, J. et al. 2006. Novel methods improve prediction of
species’ distributions from occurrence data. – Ecography 29:
129–151.
Elith, J. et al. 2011. A statistical explanation of MaxEnt for ecolo-
gists. – Divers. Distrib. 17: 43–57.
Environmental Systems Research Institute 2009. ArcGIS 9.3.1
– Geographical information system. – Environment System
Research. Inst..
Fielding, A. H. and Bell, J. F. 1997. A review of methods for the
assessment of prediction errors in conservation presence/
absence models. – Environ. Conserv. 24: 38–49.
Forero-Medina, G. et al. 2011. Constraints to species’
elevational range shifts as climate changes. – Conserv. Biol. 25:
163–171.
Fouquet, A. et al. 2010. Using ecological niche modelling to infer
past, present and future environmental suitability for Lei-
opelma hochstetteri, an endangered New Zealand native frog.
– Biol. Conserv. 143: 1375–1384.
Freeman, E. A. and Moisen, G. 2008a. PresenceAbsence: an R
package for presence absence analysis. – J. Stat. Software 23:
1–31.
Freeman, E. A. and Moisen, G. G. 2008b. A comparison of the
performance of threshold criteria for binary classification in
terms of predicted prevalence and kappa. – Ecol. Modell. 217:
48–58.
Godfrey, G. A. and Marshall, W. H. 1969. Brood break-up
and dispersal of ruffed grouse. – J. Wildlife Manage. 33:
609–620.
Guisan, A. and uiller, W. 2005. Predicting species distribution:
offering more than simple habitat models. – Ecol. Lett. 8:
993–1009.
Hawkins, B. A. et al. 2003. Energy, water, and broad-scale
geographic patterns of species richness. – Ecology 84:
3105–3117.
Hijmans, R. J. et al. 2005. Very high resolution interpolated cli-
mate surfaces for global land areas. – Int. J. Climatol. 25:
1965–1978.
Hijmans, R. J. et al. 2010. dismo: species distribution modeling.
– R package ver. 0.5-4. Available from: http://CRAN.R-
project.org/package dismo.
Hole, D. G. et al. 2009. Projected impacts of climate change on
a continent-wide protected area network. – Ecol. Lett. 12:
420–431.
Hu, J. and Jiang, Z. 2011. Climate change hastens the conservation
urgency of an endangered ungulate. – PLoS ONE 6:
e22873.
Hughes, L. 2000. Biological consequences of global warming: is
the signal already apparent? – Trends Ecol. Evol. 15: 56–61.
Hunter, C. M. et al. 2010. Climate change threatens polar bear
populations: a stochastic demographic analysis. – Ecology 91:
2883–2897.
82
omas, C. D. et al. 2004. Extinction risk from climate change.
– Nature 427: 145–148.
uiller, W. 2003. BIOMOD–optimizing predictions of
species distributions and projecting potential future
shifts under global change. – Global Change Biol. 9: 1353–
1362.
Waite, T. A. and Strickland, D. 2006. Climate change and
the demographic demise of a hoarding bird living on the edge.
– Proc. R. Soc. B 273: 2809–2813.
Walther, G. R. et al. 2002. Ecological responses to recent climate
change. – Nature 416: 389–395.
Wang, J. et al. 2011. Breeding biology of the snowy-cheeked
laughingthrush (Garrulax sukatschewi). – Wilson J. Ornithol.
123: 146–150.
Wilson, R. J. et al. 2007. An elevational shift in butterfly species
richness and composition accompanying recent climate change.
– Global Change Biol. 13: 1873–1887.
Wisz, M. S. et al. 2008. Effects of sample size on the performance
of species distribution models. – Divers. Distrib. 14:
763–773.
Zhang, X.-C. and Liu, W.-Z. 2005. Simulating potential response
of hydrology, soil erosion, and crop productivity to climate
change in Changwu tableland region on the Loess Plateau of
China. – Agric. For. Meteorol. 131: 127–142.
Zimbres, B. Q. C. et al. 2012. Range shifts under climate change
and the role of protected areas for armadillos and anteaters.
– Biol. Conserv. 152: 53–61.
Peterson, A. T. et al. 2006. Geographic potential for outbreaks of
Marburg hemorrhagic fever. – Am. J. Trop. Med. Hygiene 75:
9–15.
Phillips, S. J. et al. 2006. Maximum entropy modeling of species
geographic distributions. – Ecol. Modell. 190: 231–259.
Pounds, J. A. et al. 1999. Biological response to climate change on
a tropical mountain. – Nature 398: 611–615.
Ramirez, J. and Jarvis, A. 2008. High resolution statistically
downscaled future climate surfaces. – Centre for Tropical
Agriculture, CIAT.
Rushton, S. et al. 2004. New paradigms for modelling species dis-
tributions? – J. Appl. Ecol. 41: 193–200.
Sekercioglu, C. H. et al. 2007. Climate change, elevational range
shifts, and bird extinctions. – Conserv. Biol. 22: 140–150.
Solomon, S. et al. (eds) 2007. Contribution of working group I to
the 4th assessment report of the intergovernmental panel on
climate change. – Cambridge Univ. Press.
Storch, I. 2000. Status survey and conservation action plan
2000–2004: grouse. – IUCN, World Pheasant Association,
Gland, Switzerland.
Sun, H. L. 1996. Formation and evolution of Qinghai–Xizang
plateau – Shanghai Sci. Technol. Press, Shanghai, China.
Sun, Y. H. et al. 2003. Population ecology of the Chinese grouse,
Bonasa sewerzowi, in a fragmented landscape. – Biol. Conserv.
110: 177–184.
omas, C. D. and Lennon, J. J. 1999. Birds extend their ranges
northwards. – Nature 399: 213–213.
Supplementary material (available online as Appendix
wb-13-024 at www. www.wildlifebiology.org/readers/
appendix.). Appendix 1.
